Slides 2024

Lecture 01   Introduction to online learning

• Follow the regularized Leader (FTRL) & Online Mirror Descent (OMD)
• A special cases: Hedge Algorithm, online gradient descent
• Lower bounds for online learning problems

Lecture 02 & Lecture 03  Learning in games

• Introduction to game theory
• Coarse correlated equilibrium
• Optimistic methods in games
• Special cases: optimistic hedge and friends

Lecture 04  Online learning with bandit feedback

• Introduction to bandit feedback
• Bandit online learning algorithms
• Lower bounds for bandit online learning algorithms

Lecture 05  Variational inequalities

• An operator view of minimax problems and the associated solution concepts
• Gradient descent-ascent based on FTRL
• Gradient descent-ascent under co-coercivity
• Extragradient (EG)

Lecture 06 Variational inequalities

• EG friends: Single-call variants
• Error bounds and faster rates
• Last iterate analysis of EG